• 検索結果がありません。

Automorphisms of free groups. I — erratum

N/A
N/A
Protected

Academic year: 2022

シェア "Automorphisms of free groups. I — erratum"

Copied!
3
0
0

読み込み中.... (全文を見る)

全文

(1)

New York Journal of Mathematics

New York J. Math. 22(2016) 1135–1137.

Automorphisms of free groups. I — erratum

Laurent Bartholdi

Abstract. I report an error in Theorem A of Automorphisms of free groups. I, New York J. Math. 19(2013), 395–421, where it was claimed that two filtrations of the group of IA automorphisms of a free group coincide up to torsion.

In fact, using a recent result by Day and Putman, I show that, for a free group of rank 3, the opposite conclusion holds, namely that the two series differ rationally.

1. Introduction

Let F denote a free group of rank r. Filter F by its lower central series (Fn)n≥1, defined by F1 = F and Fn = [F, Fn−1]. Let A denote the au- tomorphism group of F, and let A1 denote the kernel of the natural map A → GLr(Z) = Aut(F/F0). The group A1 has two natural filtrations: on the one hand, its lower central series, defined as above by γ1 = A1 and γn = [A1, γn−1], and on the other handAn = ker(A1 Aut(F/Fn+1)). We have γn≤An for alln.

Andreadakis conjectures [1, page 253] thatAnn, and proves his asser- tion for r = 3, n≤3 and for r= 2. This is further developed by Pettet [7], who proves that γ3 has finite index in A3 for all r, building her work on Johnson’s homomorphism [6].

It was noted in [3, Theorem A] that, if r = 3, the groups γ7 and A7

differ, disproving Andreadakis’s conjecture. It was however also erroneously claimed there that Ann is a finite group for all n. The “proof” relied on the unproven assertion that the filtrations (γn)n≥1 and (An)n≥1 define the same topology on A1. Theorem A should be replaced by the following statement:

Theorem A. The filtrations(γn)n≥1 and(An)n≥1 differ rationally atn= 4 for r = 3, and we have

(A44)⊗Q∼=Q3.

Received September 20, 2016.

2010Mathematics Subject Classification. 20E36, 20F28, 20E05, 20F40.

Key words and phrases. Lie algebra; Automorphism groups; Lower central series.

ISSN 1076-9803/2016

1135

(2)

1136 LAURENT BARTHOLDI

Let Ac1 = proj limA1/An denote the completion ofA1 under the filtration (An)n≥1, let (cγn)n≥1 denote its closed lower central series, and let (Acn)n≥1

denote the closure of (An)n≥1 in Ac1. Then Ac7/γb7 ∼=Z/3.

Proof. In [8], Day and Putman give explicit presentations ofA1for allr, by generators, relators and endomorphisms (see [2]). Here is a small adaptation of their result. LetE be the free group generated by the set

S:={Mi,[j,k]: 1≤j6=i6=k≤r, j < k} ∪ {Ci,j : 1≤i6=j≤r}.

These are the Magnus generators of A1, and act on F respectively by xi 7→xi[xj, xk] and xi 7→xxij,

all other generators being fixed.

Day and Putman give explicit finite setsR ⊂E0 (of size around 30) and Θ⊂End(E) (of size around 4) such that

A1 ∼=hS |wθ for allw∈R and allθ∈Θi.

Furthermore, Θ induces automorphisms ofA1 that generate the conjugation action of Aut(F) on A1.

Using the algorithm described in [4], implemented in [5], it is possible to compute nilpotents quotients of A1 of arbitrary class. I entered Day and Putman’s presentation inGapforr= 3, and computed (in about 1 minute) its class-4 quotient. The result, atop the calculations in [3] gives (with ab for (Z/aZ)b)

n= 1 2 3 4

γnn+1 Z9 Z18 Z43×214×39 Z123×250×43×83×345×99 An/An+1 Z9 Z18 Z43 Z120

.

We deduce A55∼=Z3×torsion.

For the second claim, it suffices to note that the computer calculations described in [3] actually manipulate (approximations of) the groupAc1rather

thanA1.

Acknowledgments

I am grateful to Matt Day and Andy Putman for their generous insights, discussions and patience in resolving the discrepancy between their work and the original Theorem A.

References

[1] Andreadakis, Stylianos. On the automorphisms of free groups and free nilpotent groups.Proc. London Math. Soc.(3)15(1965), 239–268,MR0188307(32 #5746),Zbl 0135.04502.

[2] Bartholdi, Laurent. Endomorphic presentations of branch groups. J. Alge- bra 268 (2003), no. 2, 419–443. MR2009317 (2004h:20044), Zbl 1044.20015, arXiv:math/0007062, doi:10.1016/S0021-8693(03)00268-0.

(3)

REFERENCES 1137

[3] Bartholdi, Laurent.Automorphisms of free groups. I.New York J. Math19(2013), 395–421.MR3084710,Zbl 1288.20039,arXiv:math/1304.0498.

[4] Bartholdi, Laurent; Eick, Bettina; Hartung, Ren´e. A nilpotent quotient algo- rithm for certain infinitely presented groups and its applications,Internat. J. Algebra Comput. 18 (2008), no. 8, 1321–1344. MR2483125 (2010h:20002), Zbl 1173.20023, arXiv:0706.3131, doi:10.1142/S0218196708004871.

[5] Hartung, Ren´e. LPRES — L-Presented Groups, Version 0.3.0. 2016. http://

laurentbartholdi.github.io/lpres/.

[6] Johnson, Dennis. An abelian quotient of the mapping class group Ig.Math. Ann.

249(1980), no. 3, 225–242.MR579103(82a:57008),Zbl 0409.57009.

[7] Pettet, Alexandra. The Johnson homomorphism and the second cohomology of IAn. Algebr. Geom. Topol. 5 (2005), 725–740. MR2153110 (2006j:20050), Zbl 1085.20016,arXiv:math/0501053.

[8] Day, Matthew B.; Putman, Andrew. On the second homology group of the Torelli subgroup of Aut(Fn).arXiv:1408.6242.

[email protected]

http://www.uni-math.gwdg.de/laurent

(Laurent Bartholdi)epartement de Math´ematiques et Applications, ´Ecole Nor- male Sup´erieure, Paris and Mathematisches Institut, Georg-August Univer- sit¨at zu G¨ottingen

This paper is available via http://nyjm.albany.edu/j/2016/22-52.html.

参照

関連したドキュメント

CG1 : Carrier gas line ; CG2 : Seal gas line ; SI : A part of the conventional septum type injector ; → : Flow direction of the carrier

Recently, it was reported that ketoconazole, which is a well-known inhibitor of CYP3A4, potently inhibits the morphine glucuronosyltransferase activity catalyzed by recombinant UGT2B7

Note that Theorem 1.1 deals with the speial ase when H is the power set 2 [n] , whih is the simplest example of a hereditary family.. It is easy to see that a

If G is a finitely generated group that is residually torsion-free nilpotent, then the natural action of Aut(G) on LO(G) is faithful.. Our modifications of Koberda’s theorem replace

We show that if the fundamental group of the complement of a rationally homologically fibered knot in a rational homology 3- sphere is bi-orderable, then its Alexander polynomial has

We note that this topos is Boolean, so it does not provide a counterexample to the assertion that every completely distributive Grothendieck topos has initial normal covers for all

We will draw connections between the structure of orderings on residually torsion-free nilpotent, hyper- bolic groups and their Gromov boundaries, and we show that in those cases

Suppose the group A possesses a nice subgroup G of countable length equipped with a valuation produced by the restricted height valuation on A such that A/G is a weakly n-summable